[PATCH v3 01/10] cpufreq: arm_big_little: add cluster regulator support
Bartlomiej Zolnierkiewicz
b.zolnierkie at samsung.com
Fri Dec 4 09:30:26 PST 2015
Add cluster regulator support as a preparation to adding
generic arm_big_little_dt cpufreq_dt driver support for
ODROID-XU3 board. This allows arm_big_little[_dt] driver
to set not only the frequency but also the voltage (which
is obtained from operating point's voltage value) for CPU
clusters.
Cc: Kukjin Kim <kgene.kim at samsung.com>
Cc: Doug Anderson <dianders at chromium.org>
Cc: Javier Martinez Canillas <javier at osg.samsung.com>
Cc: Andreas Faerber <afaerber at suse.de>
Cc: Sachin Kamat <sachin.kamat at linaro.org>
Cc: Thomas Abraham <thomas.ab at samsung.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <b.zolnierkie at samsung.com>
---
.../bindings/cpufreq/arm_big_little_dt.txt | 4 +
drivers/cpufreq/arm_big_little.c | 175 +++++++++++++++++----
2 files changed, 151 insertions(+), 28 deletions(-)
diff --git a/Documentation/devicetree/bindings/cpufreq/arm_big_little_dt.txt b/Documentation/devicetree/bindings/cpufreq/arm_big_little_dt.txt
index 0715695..8ca4a12 100644
--- a/Documentation/devicetree/bindings/cpufreq/arm_big_little_dt.txt
+++ b/Documentation/devicetree/bindings/cpufreq/arm_big_little_dt.txt
@@ -18,6 +18,10 @@ Required properties:
Optional properties:
- clock-latency: Specify the possible maximum transition latency for clock,
in unit of nanoseconds.
+- cpu-cluster.0-supply: Provides the regulator node supplying voltage to CPU
+ cluster 0.
+- cpu-cluster.1-supply: Provides the regulator node supplying voltage to CPU
+ cluster 1.
Examples:
diff --git a/drivers/cpufreq/arm_big_little.c b/drivers/cpufreq/arm_big_little.c
index c5d256c..8b05137 100644
--- a/drivers/cpufreq/arm_big_little.c
+++ b/drivers/cpufreq/arm_big_little.c
@@ -31,6 +31,7 @@
#include <linux/slab.h>
#include <linux/topology.h>
#include <linux/types.h>
+#include <linux/regulator/consumer.h>
#include "arm_big_little.h"
@@ -57,6 +58,9 @@ static bool bL_switching_enabled;
static struct cpufreq_arm_bL_ops *arm_bL_ops;
static struct clk *clk[MAX_CLUSTERS];
+static struct regulator *reg[MAX_CLUSTERS];
+static struct device *cpu_devs[MAX_CLUSTERS];
+static int transition_latencies[MAX_CLUSTERS];
static struct cpufreq_frequency_table *freq_table[MAX_CLUSTERS + 1];
static atomic_t cluster_usage[MAX_CLUSTERS + 1];
@@ -125,30 +129,53 @@ static unsigned int bL_cpufreq_get_rate(unsigned int cpu)
}
}
-static unsigned int
-bL_cpufreq_set_rate(u32 cpu, u32 old_cluster, u32 new_cluster, u32 rate)
+static int
+bL_cpufreq_set_rate_cluster(u32 cpu, u32 cluster, u32 new_rate)
{
- u32 new_rate, prev_rate;
+ unsigned long volt = 0, volt_old = 0;
+ long freq_Hz;
+ u32 old_rate;
int ret;
- bool bLs = is_bL_switching_enabled();
- mutex_lock(&cluster_lock[new_cluster]);
+ freq_Hz = new_rate * 1000;
+ old_rate = clk_get_rate(clk[cluster]) / 1000;
- if (bLs) {
- prev_rate = per_cpu(cpu_last_req_freq, cpu);
- per_cpu(cpu_last_req_freq, cpu) = rate;
- per_cpu(physical_cluster, cpu) = new_cluster;
+ if (!IS_ERR(reg[cluster])) {
+ struct dev_pm_opp *opp;
+ unsigned long opp_freq;
- new_rate = find_cluster_maxfreq(new_cluster);
- new_rate = ACTUAL_FREQ(new_cluster, new_rate);
- } else {
- new_rate = rate;
+ rcu_read_lock();
+ opp = dev_pm_opp_find_freq_ceil(cpu_devs[cluster], &freq_Hz);
+ if (IS_ERR(opp)) {
+ rcu_read_unlock();
+ pr_err("%s: cpu %d, cluster: %d, failed to find OPP for %ld\n",
+ __func__, cpu, cluster, freq_Hz);
+ return PTR_ERR(opp);
+ }
+ volt = dev_pm_opp_get_voltage(opp);
+ opp_freq = dev_pm_opp_get_freq(opp);
+ rcu_read_unlock();
+ volt_old = regulator_get_voltage(reg[cluster]);
+ pr_debug("%s: cpu %d, cluster: %d, Found OPP: %ld kHz, %ld uV\n",
+ __func__, cpu, cluster, opp_freq / 1000, volt);
}
- pr_debug("%s: cpu: %d, old cluster: %d, new cluster: %d, freq: %d\n",
- __func__, cpu, old_cluster, new_cluster, new_rate);
+ pr_debug("%s: cpu %d, cluster: %d, %u MHz, %ld mV --> %u MHz, %ld mV\n",
+ __func__, cpu, cluster,
+ old_rate / 1000, (volt_old > 0) ? volt_old / 1000 : -1,
+ new_rate / 1000, volt ? volt / 1000 : -1);
- ret = clk_set_rate(clk[new_cluster], new_rate * 1000);
+ /* scaling up? scale voltage before frequency */
+ if (!IS_ERR(reg[cluster]) && new_rate > old_rate) {
+ ret = regulator_set_voltage_tol(reg[cluster], volt, 0);
+ if (ret) {
+ pr_err("%s: cpu: %d, cluster: %d, failed to scale voltage up: %d\n",
+ __func__, cpu, cluster, ret);
+ return ret;
+ }
+ }
+
+ ret = clk_set_rate(clk[cluster], new_rate * 1000);
if (!ret) {
/*
* FIXME: clk_set_rate hasn't returned an error here however it
@@ -158,25 +185,66 @@ bL_cpufreq_set_rate(u32 cpu, u32 old_cluster, u32 new_cluster, u32 rate)
* problem we will read back the clock rate and check it is
* correct. This needs to be removed once clk core is fixed.
*/
- if (clk_get_rate(clk[new_cluster]) != new_rate * 1000)
+ if (clk_get_rate(clk[cluster]) != new_rate * 1000)
ret = -EIO;
}
if (WARN_ON(ret)) {
- pr_err("clk_set_rate failed: %d, new cluster: %d\n", ret,
- new_cluster);
- if (bLs) {
- per_cpu(cpu_last_req_freq, cpu) = prev_rate;
- per_cpu(physical_cluster, cpu) = old_cluster;
+ pr_err("%s: clk_set_rate failed: %d, cluster: %d\n",
+ __func__, cluster, ret);
+ if (!IS_ERR(reg[cluster]) && volt_old > 0)
+ regulator_set_voltage_tol(reg[cluster], volt_old, 0);
+ return ret;
+ }
+
+ /* scaling down? scale voltage after frequency */
+ if (!IS_ERR(reg[cluster]) && new_rate < old_rate) {
+ ret = regulator_set_voltage_tol(reg[cluster], volt, 0);
+ if (ret) {
+ pr_err("%s: cpu: %d, cluster: %d, failed to scale voltage down: %d\n",
+ __func__, cpu, cluster, ret);
+ clk_set_rate(clk[cluster], old_rate * 1000);
+ return ret;
}
+ }
- mutex_unlock(&cluster_lock[new_cluster]);
+ return 0;
+}
- return ret;
+static int
+bL_cpufreq_set_rate(u32 cpu, u32 old_cluster, u32 new_cluster, u32 rate)
+{
+ u32 new_rate, prev_rate;
+ int ret;
+ bool bLs = is_bL_switching_enabled();
+
+ mutex_lock(&cluster_lock[new_cluster]);
+
+ if (bLs) {
+ prev_rate = per_cpu(cpu_last_req_freq, cpu);
+ per_cpu(cpu_last_req_freq, cpu) = rate;
+ per_cpu(physical_cluster, cpu) = new_cluster;
+
+ new_rate = find_cluster_maxfreq(new_cluster);
+ new_rate = ACTUAL_FREQ(new_cluster, new_rate);
+ } else {
+ new_rate = rate;
+ }
+
+ pr_debug("%s: cpu: %d, old cluster: %d, new cluster: %d, freq: %d\n",
+ __func__, cpu, old_cluster, new_cluster, new_rate);
+
+ ret = bL_cpufreq_set_rate_cluster(cpu, new_cluster, new_rate);
+ if (ret && bLs) {
+ per_cpu(cpu_last_req_freq, cpu) = prev_rate;
+ per_cpu(physical_cluster, cpu) = old_cluster;
}
mutex_unlock(&cluster_lock[new_cluster]);
+ if (ret)
+ return ret;
+
/* Recalc freq for old cluster when switching clusters */
if (old_cluster != new_cluster) {
pr_debug("%s: cpu: %d, old cluster: %d, new cluster: %d\n",
@@ -190,14 +258,11 @@ bL_cpufreq_set_rate(u32 cpu, u32 old_cluster, u32 new_cluster, u32 rate)
/* Set freq of old cluster if there are cpus left on it */
new_rate = find_cluster_maxfreq(old_cluster);
new_rate = ACTUAL_FREQ(old_cluster, new_rate);
-
if (new_rate) {
pr_debug("%s: Updating rate of old cluster: %d, to freq: %d\n",
__func__, old_cluster, new_rate);
- if (clk_set_rate(clk[old_cluster], new_rate * 1000))
- pr_err("%s: clk_set_rate failed: %d, old cluster: %d\n",
- __func__, ret, old_cluster);
+ bL_cpufreq_set_rate_cluster(cpu, old_cluster, new_rate);
}
mutex_unlock(&cluster_lock[old_cluster]);
}
@@ -304,6 +369,8 @@ static void _put_cluster_clk_and_freq_table(struct device *cpu_dev)
return;
clk_put(clk[cluster]);
+ if (!IS_ERR(reg[cluster]))
+ regulator_put(reg[cluster]);
dev_pm_opp_free_cpufreq_table(cpu_dev, &freq_table[cluster]);
if (arm_bL_ops->free_opp_table)
arm_bL_ops->free_opp_table(cpu_dev);
@@ -337,7 +404,9 @@ static void put_cluster_clk_and_freq_table(struct device *cpu_dev)
static int _get_cluster_clk_and_freq_table(struct device *cpu_dev)
{
+ unsigned long min_uV = ~0, max_uV = 0;
u32 cluster = raw_cpu_to_cluster(cpu_dev->id);
+ char name[14] = "cpu-cluster.";
int ret;
if (freq_table[cluster])
@@ -350,6 +419,51 @@ static int _get_cluster_clk_and_freq_table(struct device *cpu_dev)
goto out;
}
+ name[12] = cluster + '0';
+ reg[cluster] = regulator_get_optional(cpu_dev, name);
+ if (!IS_ERR(reg[cluster])) {
+ unsigned long opp_freq = 0;
+
+ dev_dbg(cpu_dev, "%s: reg: %p, cluster: %d\n",
+ __func__, reg[cluster], cluster);
+ cpu_devs[cluster] = cpu_dev;
+
+ /*
+ * Disable any OPPs where the connected regulator isn't able to
+ * provide the specified voltage and record minimum and maximum
+ * voltage levels.
+ */
+ while (1) {
+ struct dev_pm_opp *opp;
+ unsigned long opp_uV;
+
+ rcu_read_lock();
+ opp = dev_pm_opp_find_freq_ceil(cpu_dev, &opp_freq);
+ if (IS_ERR(opp)) {
+ rcu_read_unlock();
+ break;
+ }
+ opp_uV = dev_pm_opp_get_voltage(opp);
+ rcu_read_unlock();
+
+ if (regulator_is_supported_voltage(reg[cluster], opp_uV,
+ opp_uV)) {
+ if (opp_uV < min_uV)
+ min_uV = opp_uV;
+ if (opp_uV > max_uV)
+ max_uV = opp_uV;
+ } else {
+ dev_pm_opp_disable(cpu_dev, opp_freq);
+ }
+
+ opp_freq++;
+ }
+
+ ret = regulator_set_voltage_time(reg[cluster], min_uV, max_uV);
+ if (ret > 0)
+ transition_latencies[cluster] = ret * 1000;
+ }
+
ret = dev_pm_opp_init_cpufreq_table(cpu_dev, &freq_table[cluster]);
if (ret) {
dev_err(cpu_dev, "%s: failed to init cpufreq table, cpu: %d, err: %d\n",
@@ -483,6 +597,11 @@ static int bL_cpufreq_init(struct cpufreq_policy *policy)
else
policy->cpuinfo.transition_latency = CPUFREQ_ETERNAL;
+ if (cur_cluster < MAX_CLUSTERS &&
+ policy->cpuinfo.transition_latency != CPUFREQ_ETERNAL)
+ policy->cpuinfo.transition_latency
+ += transition_latencies[cur_cluster];
+
if (is_bL_switching_enabled())
per_cpu(cpu_last_req_freq, policy->cpu) = clk_get_cpu_rate(policy->cpu);
--
1.9.1
More information about the linux-arm-kernel
mailing list